gensai.aiCompliance assistant on sovereign RAG
Client chatbot to query a protected legal knowledge base
Project
A sovereign chatbot was deployed for a compliance firm to provide precise regulatory answers from each client's own knowledge base. Thanks to a personalized RAG, the AI queries internal documentation while taking the client's industry and company size into account. This solution frees up time for legal experts, who can focus on complex, high-value topics.
The need
The clients of a compliance firm ask regulatory questions whose answer depends on their sector and their size. Answering them ties up legal experts on requests that are often similar to one another.
How it works
The assistant answers from each client's own knowledge base, hosted in a sovereign environment at Scaleway, with a Mistral model. It runs at a low temperature to give consistent answers from one question to the next, and passes complex requests on to the right expert at the firm.
What the project shows
A regulatory assistant is only valuable if it stays within its sources. The RAG architecture anchors every answer in the firm's documents, and the expert remains the fallback as soon as a question goes beyond that scope.